[SciPy-dev] [SciPy-user] eig() segfaults on SuSE 9.3 with ACML, Numeric's eigenvectors works

Arnd Baecker arnd.baecker at web.de
Mon Nov 14 04:58:13 CST 2005


On Mon, 14 Nov 2005, Travis Oliphant wrote:

> Arnd Baecker wrote:
>
> >Program received signal SIGSEGV, Segmentation fault.
> >[Switching to Thread 46912507335168 (LWP 31397)]
> >0x00002aaaab3a2cdb in memmove () from /lib64/tls/libc.so.6
> >(gdb) bt
> >#0  0x00002aaaab3a2cdb in memmove () from /lib64/tls/libc.so.6
> >#1  0x00002aaaab96246c in PyArray_CopyInto (dest=0x2aaab55609e0, src=0x8)
> >at arrayobject.c:658
> >#2  0x00002aaaab96735b in array_imag_set (self=0x2aaab55608f0,
> >val=0x83000d690097dd63) at arrayobject.c:4260
> >
> >
> Yes, thanks for the trace back.  I can't believe that one got through
> the tests.  It's a pretty obvious error: (adding to the strides vector
> instead of the data).

Great that you found it so quickly!

The example works fine now with
In [3]: scipy.__core_version__
Out[3]: '0.6.2.1483'

> I guess it just emphasizes the need for more tests.

Looks like - so maybe you better wait for the 1.0
a bit longer (so that it appears as rock-solid as
possible from the very beginning), but encourage more testing
in terms of converting bigger projects to newcore ...

Best, Arnd

P.S.: thanx to the one who has put fftw3 support into newscipy!




More information about the Scipy-dev mailing list